The present invention deals with the handling of disadvantages of a communication protocol. Such disadvantages (as it will be seen later) are inherent to many types of communication protocols. As wireless communication protocols are the most prominent and well-known protocols reflecting such disadvantages, the current description will concentrate on wireless communication protocols. Nevertheless, the same or similar disadvantages may be found in other types of communication protocols; of course the present invention may be applied to these communication protocols as well. The invention is independent from the communication protocol it is applied to.
For several years industry watchers have been forecasting an explosion in wireless Internet usage. With more than 200 million Internet users and more than 400 million mobile subscribers in the marketplace there is every reason to believe that this industry, as it grows, will have a dramatic impact on the way we access information.
The mobile communications marketplace continues to expand explosively, with potential revenue growth supported by an ever-increasing variety of new services and new target segments. Each of these new services and segments bring with it fresh challenges for business activities outside traditional office settings. The wireless Internet allows businesses to deliver new types of services, including both internal services like sales automation and document management; and external services like travel reservations, stock trading, information selling—faster and easier than ever before. Mobile data communication will set new business standards for timely access to people and information. Managers, business partners, and account executives-all of whom are expected to spend more of their time in the field-will profit from remote access to enterprise networks.
Mobile data communication and its increasing acceptance by users will substantially influence the advancement of the terrestrial network. Apart from the infrastructural effects, which result from the spreading of mobile radio data transmission networks, there are special problems such as:
1. Low bandwidth and transmission speed—the transmission bandwidth of radio data transmission services remains far behind that of stationary networks.
2. High costs—transmission costs over wireless communication networks are much higher than the costs over stationary networks.
3. High complexity—in a dynamic architecture, logical connections must be mapped on different physical structures.
4. Low reliability—wireless connections are significantly less reliable then wireline connections.
5. High latency—the response time for wireless links is much slower than that of terrestrial links.
6. High connection overhead—each data request for a TCP/IP based server requires the client part to open an TCP/IP socket. The consequences are to intensify the data overhead and to increase the latency.
A wide variety of problems may arise up when wireless communication terminals send and receive signals over the air. The signals of all the terminals are subject to mutual interference. The characteristics of the propagation medium change randomly as users move, and the mobile radio channel introduces random variation in the received signal power and other distortions, such as frequency shifts and the spreading of signals over time. Signals that travel over the air are also more vulnerable to jamming and interception than are those transmitted through wires or fibers. As a result, transmitted data packets may be lost. These limitations are often addressed with a combination of sophisticated signal processing techniques and antennas, but there is no comprehensive software based solution. However, these solutions add to the complexity of wireless networks and increase power requirements.
Many of these shortcomings may be attributed to limited bandwidth, which additionally drives up the costs of wireless data links. At present, transmission speed is limited to 9600 Bit/s. This limitation is inherent to GSM (Global System for Mobile communication), which transfers only approximately 13 kBit/s per channel. There have been attempts to moderate these disadvantages. The GSM phase II standard specified a data mode supporting 14.4 kBit/s. But the increased rate comes at the expense of Forward Error Correction, and therefore lowers the quality of the connections. Fundamentally GSM was not designed for mobile Internet access, and even a 14.4 kBit/s data rate offers only a small improvement.
In contrast to other areas of information technology, wireless communications has yet to converge to a single technical standard or even a very small number of them. Instead it appears that diversity will endure for the foreseeable future. As long as this technical standard is not available, other ways must be investigated (perhaps based on software solutions), to provide a solution to the above problems.
A further dimension of the problem is introduced by the use of TCP/IP over wireless networks. Such a combination of a first and a second protocol is sometimes inevitable. The first protocol, TCP/IP, has to be used because it is simply the established protocol of the Internet; on the other hand the second protocol, the wireless communication protocol, must be used due to the specific communication environment for which there is no other protocol available for substitution. In such situations one might be confronted with the problem of how to deal with disadvantageous characteristics of a certain protocol which either might be inherent to the protocol itself or which might be the result of combining two protocols.
In the current situation of TCP/IP over wireless networks, high delay and variation in data loss result in unacceptable performance for many standard multimedia applications and reliable protocols such as TCP/IP. Both multimedia applications and reliable protocols adapt to long term end-to-end estimates of delay and packet loss between the data source and destination. However, they do not perform well when rapid variations in network characteristics occur, causing high fluctuations in these estimates. In order for these applications and protocols to achieve good performance, the protocol for transmitting data to mobile hosts must provide communication with reliable connections and negligible data loss (which is not the case for wireless communication protocols).
Typically, wireless Internet access works in the same way as network access using fixed data modems. Usually the mobile terminal (a combination of Notebook and wireless data phone) calls a fixed network modem placed on the ISP (Internet Service Provider) side. Thereby it make use of the Point-to-Point (PPP) (RFC1662) or the SLIP(RFC1055) in order to enable TCP/IP over phone lines (additionally there are proprietary solutions of individual portable radio network carriers). But both PPP and SLIP are not very well suited for unreliable radio connections because of transmission overhead. There is a certain amount of transmission overhead associated with maintaining timers, scheduling processes, and specific protocol control data.
IP (Internet Protocol) is a connectionless packet-oriented protocol of the network layer of the OSI reference model. In the transport layer usually TCP (transport control protocol) is applied. TCP uses IP. TCP is a connection-oriented and reliable protocol, including error recognition and correction, flow control, avoidance of congestion in routers, and fairness among network components. The TCP protocol transfers data complete and without errors. The price is slower transmission over error susceptible channels. But by using a perfected windowing technique, TCP minimizes this price. A sliding window allows TCP to send several data segments and await their acknowledgment. As soon as an acknowledgment is received, the window is shifted and another segment can be sent. For every sent segment TCP starts a separate timer, which possibly signals a missing acknowledgment and initiates a retransmission of the segment. With the help of Congestion Avoidance, Multiplicative Decrease, and Slow Start, TCP adapts to the network condition and avoids an overload of the network.
Today's networks offer very low error rates (˜10-6). The TCP mechanisms are therefore designed for wired networks with low error rates. A typical wireless network can't provide such good transmission quality and small delays. Moreover the lower OSI layers for wireless networks use techniques for error recognition and correction, which increase the delays. TCP may interpret such delays as evidence of congestion. While the perfected mechanisms of the wireless network layers provide a faultless transmission, TCP timers expire and initiate retransmissions. These timers are adapted dynamically by measuring the round trip time. A new time is only taken when an acknowledgment is received for a segment, which has not yet been retransmitted. After a period of error-free transmission the timers are accordingly short.
If there is a short phase of disturbance or poor transmission conditions, however, the error correction mechanisms of the wireless network layers cause longer delays and thereby longer TCP round trip times. TCP reacts with expiring timers and unnecessary retransmissions. TCP always interprets expired times (or data loss) as a sign of congestion. The effects are longer timers and a reduction of size of the sliding window which causes a drop in transmission rate. Even when the transmission in the network recovers, TCP still needs some time to adapt its timers to this condition.
Since TCP interprets all acknowledge delays as congestion, it can't react correctly in these situations which are typical to a wireless network. So TCP is not the optimal protocol for transferring data in wireless networks. Thus there is a need for a way of compensating for disadvantageous characteristics of a communication protocol in situations where the communication protocol itself cannot be replaced.